Sql profiles用法介绍

来源：互联网发布：淘宝卖家收到法院传票编辑：程序博客网时间：2024/05/21 17:02

DBMS_SQLTUNE优化SQL 是在oracle10g才出来的新特性，使用它能很大程度上方便对sql的分析和优化。执行DBMS_SQLTUNE包进行sql优化需要有advisor的权限：

grant advisor to user;

使用DBMS_SQLTUNE包进行SQL优化，大致可以分为四个步骤：

创建优化任务
执行优化任务
显示优化任务的结果
按照建议执行优化

以下以sys用户为例，使用SQL Profile过程: 以下程序摘自（Oracle DBA手记2 page94-98）

1.创建测试表及索引
SQL> create table big_table_prof as select rownum
2 as id, t.* from all_objects t;

表已创建。

SQL> create index idx_big_table_prof on big_table_prof(id);

索引已创建。

2.查看执行计划

2.1 分析表big_table_prof
SQL> exec dbms_stats.gather_table_stats(user,'big_table_prof');

PL/SQL 过程已成功完成。

SQL> set linesize 100
SQL> set autotrace traceonly exp

2.2 访问索引SQL
SQL> select count(*) from big_table_prof where id<100;

执行计划
----------------------------------------------------------
Plan hash value: 47754723

----------------------------------------------------------------------------------------
| Id | Operation         | Name               | Rows | Bytes | Cost (%CPU)| Time     |
----------------------------------------------------------------------------------------
|   0 | SELECT STATEMENT |                    |     1 |     5 |     2   (0)| 00:00:01 |
|   1 | SORT AGGREGATE   |                    |     1 |     5 |            |          |
|* 2 |   INDEX RANGE SCAN| IDX_BIG_TABLE_PROF |    62 |   310 |     2   (0)| 00:00:01 |
----------------------------------------------------------------------------------------

Predicate Information (identified by operation id):
---------------------------------------------------

2 - access("ID"<100)

2.3 强行加入full hints使其变为全表扫描：
SQL> select /*+full(t)*/ count(*) from big_table_prof t where id < 100;

执行计划
----------------------------------------------------------
Plan hash value: 3149272518

-------------------------------------------------------------------------------------
| Id | Operation          | Name           | Rows | Bytes | Cost (%CPU)| Time     |
-------------------------------------------------------------------------------------
|   0 | SELECT STATEMENT   |                |     1 |     5 |   186   (2)| 00:00:03 |
|   1 | SORT AGGREGATE    |                |     1 |     5 |            |          |
|* 2 |   TABLE ACCESS FULL| BIG_TABLE_PROF |    62 |   310 |   186   (2)| 00:00:03 |
-------------------------------------------------------------------------------------

Predicate Information (identified by operation id):
---------------------------------------------------

2 - filter("ID"<100)

3.调用dbms_sqltune调优建议者以创建调优任务，并报告执行结果

1 declare
2 v_task_name varchar2(40);
3 begin
4 v_task_name := dbms_sqltune.create_tuning_task
5 (sql_text=>'select /*+full(t)*/ count(*) from big_table_prof t where id<100',
6 task_name=>'task_big_table_prof');
7 dbms_sqltune.execute_tuning_task(v_task_name);
8 dbms_output.put_line(v_task_name);
9* end;
SQL> /

PL/SQL 过程已成功完成。

4.SQL> select status from user_advisor_tasks where task_name='first_tuning_task13';

STATUS
-----------
COMPLETED

5. 显示优化任务的结果
SQL> set serveroutput on size 100000
SQL> set long 100000
SQL> select dbms_sqltune.report_tuning_task('task_big_table_prof') from dual;

DBMS_SQLTUNE.REPORT_TUNING_TASK('TASK_BIG_TABLE_PROF')
--------------------------------------------------------------------------------
GENERAL INFORMATION SECTION
-------------------------------------------------------------------------------
Tuning Task Name                  : task_big_table_prof     调优任务名
Tuning Task Owner                 : SYS                     所有者
Scope                             : COMPREHENSIVE
Time Limit(seconds)               : 1800
Completion Status                 : COMPLETED
Started at                        : 06/30/2012 17:37:31
Completed at                      : 06/30/2012 17:37:31
Number of SQL Profile Findings    : 1

DBMS_SQLTUNE.REPORT_TUNING_TASK('TASK_BIG_TABLE_PROF')
--------------------------------------------------------------------------------
-------------------------------------------------------------------------------
Schema Name: SYS
SQL ID : 65mnw6588hyjg
SQL Text : select /*+full(t)*/ count(*) from big_table_prof t where id<100

-------------------------------------------------------------------------------
FINDINGS SECTION (1 finding)
-------------------------------------------------------------------------------

1- SQL Profile Finding (see explain plans section below)
--------------------------------------------------------

DBMS_SQLTUNE.REPORT_TUNING_TASK('TASK_BIG_TABLE_PROF')
--------------------------------------------------------------------------------
为此语句找到了性能

Recommendation (estimated benefit: 98.92%)
------------------------------------------
-考虑接受推荐的 SQL
execute dbms_sqltune.accept_sql_profile(task_name =>
'task_big_table_prof', replace => TRUE);

-------------------------------------------------------------------------------
EXPLAIN PLANS SECTION
-------------------------------------------------------------------------------

DBMS_SQLTUNE.REPORT_TUNING_TASK('TASK_BIG_TABLE_PROF')
--------------------------------------------------------------------------------

1- Original With Adjusted Cost
------------------------------
Plan hash value: 3149272518

--------------------------------------------------------------------------------
| Id | Operation          | Name           | Rows | Bytes | Cost (%CPU)| Time
--------------------------------------------------------------------------------
|   0 | SELECT STATEMENT   |                |     1 |     5 |   186   (2)| 00:00
3 |
|   1 | SORT AGGREGATE    |                |     1 |     5 |            |

DBMS_SQLTUNE.REPORT_TUNING_TASK('TASK_BIG_TABLE_PROF')
--------------------------------------------------------------------------------
|* 2 | TABLE ACCESS FULL| BIG_TABLE_PROF | 62 | 310 | 186 (2)| 00:00
--------------------------------------------------------------------------------

Predicate Information (identified by operation id):
---------------------------------------------------

2 - filter("ID"<100)

2- Using SQL Profile
--------------------
Plan hash value: 47754723

DBMS_SQLTUNE.REPORT_TUNING_TASK('TASK_BIG_TABLE_PROF')
--------------------------------------------------------------------------------

--------------------------------------------------------------------------------
| Id | Operation         | Name               | Rows | Bytes | Cost (%CPU)| Ti
|
--------------------------------------------------------------------------------
|   0 | SELECT STATEMENT |                    |     1 |     5 |     2   (0)| 00
|   1 | SORT AGGREGATE   |                    |     1 |     5 |            |
|* 2 |   INDEX RANGE SCAN| IDX_BIG_TABLE_PROF |    62 |   310 |     2   (0)| 00
--------------------------------------------------------------------------------
------

DBMS_SQLTUNE.REPORT_TUNING_TASK('TASK_BIG_TABLE_PROF')
--------------------------------------------------------------------------------
Predicate Information (identified by operation id):
---------------------------------------------------

2 - access("ID"<100)

-------------------------------------------------------------------------------

6.按照建议执行优化

SQL> set linesize 100
SQL> select /*+full(t)*/ count(*) from big_table_prof t where id < 100;

COUNT(*)
----------
99

执行计划
----------------------------------------------------------
Plan hash value: 47754723

Predicate Information (identified by operation id):
---------------------------------------------------

2 - access("ID"<100)

Note
-----
- SQL profile "SYS_SQLPROF_014e0f6d696a4000" used for this statement [此处注明此SQL语句使用了SQL profile优化].

由此可知，当我们无法改变应用程序的SQL语句时，可以利用SQL Profiles来影响执行计划。

Made by DylanX.